« July 2007 | Main | October 2007 »

September 23, 2007

A Unified Fabric for the Data Center?

The holy grail of data center networking has been discussed for a number of years and many have attempted to design a single technology to deliver on all of the networking requirements of data center applications. The design goal is reasonably simple: design a single data center transport that can simultaneously transmit IP and Fibre Channel traffic over a single connection. The problem is, it just isn't that simple. Data center managers expect that the transport must include sophisticated management capabilities to enable accurate depiction of what is happening within the fabric; the fabric must offer high performance and low latency, robust security; the fabric must not - absolutely ever - drop a single storage frame. Oh, and it must be able to scale to thousands – if not tens of thousands - of devices, support 10/100/1000 and 10GE attached servers, support legacy applications, be virtualizable, enable efficient utilization of IT assets, and reduce power and cooling overhead. If it can make coffee, that's a bonus. OK, the last one is a stretch, but you get the point.

It can be seen that although a number of technologies exist that could potentially address the needs of a Unified Fabric, most technologies require significant development to fully address the requirements listed above. If we take InfiniBand as an example, although it has the right performance characteristics, offers IP and Fibre Channel communications over a single interface and reduces power and cooling overhead by reducing the number of interfaces and fabric connections required to support a server, it lacks the scaling, embedded services and management capabilities that data center managers have come to expect from their Ethernet and Fibre Channel infrastructure. It also introduces one significant issue: certification against existing hardware, operating systems and applications. This latter point should not be under-estimated because even if IP-over-InfiniBand is used, hardware and software driver certification can be time consuming, costly and introduces additional complexity. To a certain extent these factors have limited adoption of the technology to high-performance computing clusters and high-performance systems such as those found on Wall Street.

One technology does however offer the promise of delivering on the promise of Unified Fabric: Ethernet. Ethernet has proven to be a survivor. It has outlasted ATM, FDDI, 100BaseVG-ANYLAN and Token Ring. It has scaled from its humble origins of shared 10Mbps, to 10Mbps switched and then on to 100Mbps, 1Gbps and 10Gbps - with 40Gbps and 100Gbps promised in the near future - without changing the frame format. Ethernet management has also evolved such that DC managers can extract information regarding traffic conditions on the network - down to individual packets if required - that assist in troubleshooting, performance monitoring, and forensic analysis. These attributes have made Ethernet the de-facto standard for the vast majority of networked devices - refrigerators and electric guitars are now available with Ethernet connections.

So, where do we go from here? There is a lot of effort going into a proposal called FCoE (Fiber Channel over Ethernet). The protocol draft (more at http://www.fcoe.com) describes a method to encapsulate a Fibre Channel frame in a regular Ethernet frame and it looks like it has wide industry support.

Certainly, this seems like an appropriate approach to solve the unification problem but this is not the first time the industry has tried to converge storage and data traffic over a single fabric. The iSCSI protocol has been around for some time but with limited success. Adoption does seem to be increasing but Fibre Channel advocates doubt its ability to deliver the performance and reliability of Fibre Channel SANs.

Are there other methods worth exploring? Is it even worth focusing so much effort on solving this problem? As one of my colleagues joked recently, "You can run fresh water and sewage in the same pipe, but why would you want to?" That might be a bit harsh but convergence of data center network fabrics would most likely lead to lower overall operational costs similar to the benefits achieved with the convergence of voice and data networks. For now, it looks like this journey will continue.

Posted by Deepak Munjal at 04:29 PM Permalink | Comments (2) | TrackBacks (0)

September 20, 2007

Who Invited You?

We were asked several times last week by several separate reports and press types some fun questions like, "Why is Cisco here at VMWare World?" and "What does Cisco have to do with Virtualization?" Some were even quite challenging which always makes for a fun day! So I figured I would put an answer together or at least a few thoughts...

Have you ever tried to move a virtual machine from one server to another? How about doing it from one rack to another?

You may find it gets hard especially if the access layer switches are in different subnets. In fact you can do it but then DNS breaks, your connections break, etc...

Have you ever had two servers that require their own security profiles? That may need to be on separate networks or segmented from each other by a firewall?

If this is the case you need to preserve network segmentation THROUGH the hypervisor yet linked with the network. Preferably in a way that doesn't do what I often refer to as 'tromboning' of traffic between different places. In a low-bandwidth or very latency sensitive environment this can be sub-optimal.

Have you ever had a broadcast storm on your network? If you move a MAC address from one port to another across different switches the upstream devices may take several minutes to clear out their CAM table timers. Until cleared traffic will either go the wrong place, or at worst go everywhere causing other devices around the data center to have to parse the traffic and determine it is not for them. This causes an interrupt and consumes bandwidth and CPU with no benefit.

There area few examples of why Cisco is partnering with VMWare and how we feel that a network approach with a server virtualization approach can benefit all of the IT professionals who work in the data center. But to put some icing on the cake let me add one more: Some day please provision a VMWare Server Guest OS for me, load a web server on it. Then without touching anything in the network bring it into production without configuring a switch, router, load balancer, or firewall ye maintain compliance with the operational procedures and governance and regulatory compliance guidelines that are necessary to run a responsible business or government entity in today's wonderfully litigious society.

Since I think the above is tacitly impossible let's agree that we need three things to execute an IT workload.

1) Something to process data - we may call these things 'Servers' :)

2) Something to store data before and after processing - 'Storage'.

3) Something to move data from where it is being stored to where it is being processed and to the humans who want to see the results of this processing - 'Networks' if I may.

Take any one away and your data center is dead. Work together, collaborate, integrate, and share resources, ideas, and operational best practices and virtualization will be an enabling technology across all three that transforms your business capabilities and efficiencies.

dg

Posted by Douglas Gourlay at 11:57 AM Permalink | Comments (0) | TrackBacks (0)

September 19, 2007

How do YOU use WAN optimization?

WAN optimization has become one of the "hot" technolgies within the networking -- and broader IT -- communities over the last 24 months. Market sizing (Gartner and others note it will be +/- $1B USD very shortly) reflects the rapid adoption of this technology.

Much of this fast adoption is due to the fact that it: 1) Addresses a very real set of problems (more on this shortly), 2) Has a very clear set of ROI calculations and/or problem resolutions (e.g. measurable application response times), and 3) Covers a range of IT challenges and sectors (branch server consolidation, application performance over WAN, bandwidth costs, data storage in branch).

So here's my question to our blog readers: do YOU use WAN optimization today? If so, exactly WHAT do you use it for? How do you measure success on your investment? Hard #'s or soft ones?

I'll throw out the first example: a large phone service provider/retailer started deploying WAAS earlier this year in their data center and retail branches across North America. They have recently found they saved 3.2 Tb (yes, that's TERAbits) of WAN traffic and related expense within a one month period. While consolidating branch infrastructure and reducing TCO from that.

Would be keen to hear your experience with your WAN optimization deployment...

Posted by Mark Weiner at 09:53 AM Permalink | Comments (2) | TrackBacks (0)

September 11, 2007

Virtualization This Week

Could Virtualization Fundamentally Alter the Computing Landscape?

I just read a recent article on BusinessWeekOnline. What jumped out at me: “Some experts believe virtualization could fundamentally alter the computing landscape as companies cope with storing and transmitting ever-growing piles of data….Companies gorged on low-priced servers to handle tasks like delivering Web pages and planning production schedules. The result: a data-center obesity epidemic, with thousands of machines running way below capacity.”

Second to maybe SOA, virtualization is the most misrepresented technology. The message that seems to be perpetuated is that it only applies to servers. Some people think for instance that the V-word is strictly VMWare or Xen. But critical areas of the overall data center like storage and networks can also be virtualized—a key concept given that storage capacity is growing faster and fatter than server capacity. The prescription that makes most sense here is a balanced diet of dynamic virtualization.

So what is Dynamic Data Center Virtualization?

When an infrastructure virtualizes the network and critical transport network services (like firewalls, load balancers, etc), server, and storage in the data center we call this Data Center Virtualization.

However, most data center virtualization scenarios today are actually static: resources in the data center are virtualized but provisioning of these resources is done manually and is not changed, added to, or moved too often. A better mid-term approach for enterprises and service providers in this case would be to add an orchestration system; something that links the elements together and allows for addition or subtraction of resources aligned to a situation based on actual traffic or, even better, business metrics, service level agreements, etc.

In a dynamically virtualized model, time is critical. The faster a re-provisioning event can occur, the more responsive the application can be to the business. When it takes minutes, the main problem solved is the elimination of human-errors and the assurance of compliance with corporate and regulatory policy. As re-provisioning time goes to 30 seconds, or even better, under 10 seconds, real-time and dynamic changes to the IT workload become more responsive, ensuring user experiences and IT service levels are maintained.

(By the way, two resources that offer good information on this topic are http://www.vmware.com/community/index.jspa and http://blogs.vmware.com/vmtn/)

From a business standpoint, virtualization of the network is important because it can drive increased efficiencies in power draw and cooling/heat dissipation in the data center (for facilities best practices and power efficiency, check out www.thegreengrid.org). It is also important because, as we all know, networks link everything together.

Are email servers used all the time? Is usage 80% or greater constantly? Do your servers peak during work-days and ebb during off hours? Why not move the instance of an email server, in real-time, to a machine with many other instances in the off hours? Then, as traffic increases, you can dynamically revert back to a dedicated machine. This would require server and network to both be virtualized and to work together--- and generate a solution that would yield greater efficiency.

Networks, servers, storage, and applications all need to be coordinated to make this vision possible. Server virtualization is a good start, but until the overall data center is virtualized and the re-provisioning times compressed, the full impact of what dynamic virtualization can enable will not be achieved.

Finally, if you’re heading to VMWorld this week, take some of this with you – I guarantee you’ll get a whole lot more out of your experience.

Posted by Douglas Gourlay at 11:12 AM Permalink | Comments (4) | TrackBacks (1)